My real OB speaker project

Discussion in 'Speakers' started by Serious, Mar 13, 2017.

Tags:
  1. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,153
    Trophy Points:
    113
    Location:
    Estonia
    Thanks @Serious and @Donald North. I do appreciate your guidance.
    In the wide internet wrt IB/OB there is not much more than home theater noobs bragging about performance they perceive with their butt rather than ears.
     
  2. Hrodulf

    Hrodulf Prohibited from acting as an MOT until year 2050

    Pyrate
    Joined:
    Sep 28, 2015
    Likes Received:
    4,276
    Trophy Points:
    113
    Sometimes it's the same thing.

    Dig the thread overall. My current reference speaker is the LX521 which is open baffle. Had I the space and funds, would totally build a pair.
     
  3. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    I've been working a lot on the crossover lately. The crossover makes or breaks the speaker. In fact I would say It is the speaker. Because (at the wavelengths involved) I couldn't get proper measurements indoors I decided to take (more or less) anechoic measurements. For this me and my dad lifted the speakers on top of our garage and pointed them facing off the garage. This way I got the speakers almost 3m off the ground. I measured at 2m distance at an angle similar to what I listen at (about 12° for this measurement if I had to guess), about 10db below 2.83Vrms, but I referenced the measurements to 2.83Vrms/1m.

    Anechoic (more or less) measurements:
    AnechoicFR.png

    I had to aggressively smooth/window the measurements to get rid of environmental noises. Reflections didn't seem to be a big issue.

    Now there is a cone resonance at 2kHz with the Voxativ driver, similar to the other measurements you'll find on the net, but once you start taking measurements in room where a lot more than the direct sound influences the FR it goes away. Much like with the stereophile measurements. The treble also looks incredibly smooth and extended in room. Subjectively the in-room measurements correlate better with what I hear. The FE83En had a more noticeable 2kHz resonance.
    The treble is a bit hot in this measurement and I probably listen a few degrees more off-axis. What looks like peaks and dips in the treble are most likely just interference effects from the big radiating surface which I don't think are as bothersome subjectively.
    The woofers roll off at a lower point because they're in a much bigger baffle. Baffle for the woofers is 60x80cm, the widebander is in a 60x30 baffle. In-room the woofers extend to the mid-30s (their Fs in the baffle is at 34.6Hz), with usable extension actually down to my lowest room mode at 25Hz.

    The sensitivity of the woofers and the Voxativ driver looks similar, but keep in mind that the Voxativ driver is probably closer to 12 Ohm vs the 8 Ohm for the woofers. In room I'm at about 98db/2.83Vrms at the listening position, but the impedance is quite nasty in the bass (in the 3 Ohm range). Definitely not tube-amp friendly.

    These measurements showed me what I had already suspected: The Voxativ driver, because of its high sensitivity, has an inherently rising FR below 1kHz that makes the slope for the rolloff slightly too steep.* This makes it almost impossible to get a proper crossover, because the phase isn't quite right.
    Right now there are two main options I see:
    1. Try to avoid cancellation in the crossover region. This is what I already tried. The drivers hardly add, but as long as they don't cancel it's fine. I tried to achieve this by comparing the CSDs (in room) in the 150-2kHz region between only Voxativ driver and with the woofers added. I adjusted the crossover until they didn't cancel anymore. There's still some slight cancellation around 2kHz (from the woofer's peak), but that might not be so much of a bad thing.
    2. Build an EQ circuit for the Voxativ driver to broadly lower the region around 500Hz to match a first order slope in the crossover region. With a bit of a higher crossover point the should add to a level that is similar to what it was before.
    3. (Completely ditch the crossover design and go with a 2nd order slope. Invert the Voxativ drivers and done! (I don't like the woofers inverted.) )
    Now, I think there's something magical about first order crossovers that you can't get with any other crossover, at least when done right. But a crossover that doesn't add is definitely not a crossover done right. Does anyone have an idea?

    *I already knew this would be a problem from the start, but the nearfield measurements I've taken didn't look too bad. Nearfield FR (measured at the edge of the whizzer) attached.



    I also measured the ultrasonic extension and there actually seemed to be a peak just below 50kHz that almost reached the 20kHz level! That's better than I expected, but really don't expect a whizzer cone to extend as high as a regular tweeter or ribbon, etc. It does however seem to reach 20kHz quite easily, at least according to my ear and in-room measurements.

    On a more serious note I also measured the bass distortion, at least I did an FFT with a 36Hz tone playing from both speakers at 87db at the listening position. That is about the lowest I feel it comfortably extends and 87db was already quite loud. This was with the H-frame which boosts the sensitivity about 2-3db below 150Hz. I didn't disconnect the widebander and I doubt it would influence the results much.
    36Hz87db.jpg
    D3 is dominant at -56db or 0.16%
    D2 is at about -66db or 0.05%
    D5 is the only other harmonic I can make out and it seems to be at about -75db or 0.018%

    Now obviously the woofers don't really have to work hard at that level, but I rarely listen that much louder. Sure, at 70dbA listening levels you can get peaks to 100db or higher, but I feel 87db is a good reference. Distortion was very low with this measurement, comparable to good headphones. It's not perfectly fair because both speakers are playing, so it might be better to compare it against 81db headphone distortion measurements. The distortion I measured at 1kHz at a slightly lower level was even lower than this, but not on the level of good headphones.
    I tend to prefer low levels of D3 over low levels of D2 (at least in the bass), but higher levels of third order distortion sound worse than higher levels of 2nd order distortion to me. The bass reminded me a bit of the UERM and I feel this explains why - D3 seems to be dominant compared to the usual 2nd order dominant dynamic headphone bass. Now that I have 2 woofers per side it actually reminds me less of the UERM - the bass got quite a bit cleaner and the distortion measurements seem to reflect this.

    I'm also trying to find the right compromise for the woofer baffle. The 50cm wide cardboard baffles that I attached to the sides of the baffle for an H-frame seemed to almost boost the bass a bit too much for me (about 2-3db). I'm thinking something in-between would be perfect, but I also want to experiment with U-frames, possibly to direct more bass towards me than towards the back. I'll try to find a compromise soon. Without any frames attached to it the bass quantity is already a bit more than the HD800, especially in the important 30-80Hz region, but it's still a bit lower than I'd like.
     

    Attached Files:

    Last edited: May 11, 2017
  4. Priidik

    Priidik MOT: Estelon

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    2,153
    Trophy Points:
    113
    Location:
    Estonia
    These are some sweet figures. Did you try different tones too?
    It would be interesting to see if 'features' in impedance graph of Betas show up.

    Which headphones? Haven't seen them much lower than 0.1%.
     
  5. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Yup, awesome bass distortion results, at least at that level. Higher frequencies should give less D3, but I haven't taken measurements.
    The impedance graphs I took didn't show any resonance wiggles, apart from the 2kHz peak, but I want to make double sure and take new measurements. I'm still a bit sceptical about cone breakup or similar effects at such low frequencies, but I haven't heard any of those more expensive woofers. Impedance bumps should also show up in nearfield FR measurements.
    The ones for the Voxativ drivers did show a small 2,3 and 8kHz wiggle.
    I got about 0.1% THD at 1kHz at 80db. With my HD800 I measured about 0.025% D2 at 95db, with D3 buried below the noise floor more than 90db down. That's much less distortion at a much higher level. Electrostatic speakers will have less midrange distortion, but most likely not lower bass distortion. Good three-way speakers will also have lower midrange and treble distortion, especially once you start to really crank it up, but then again I don't care so much about midrange distortion at 100db - that's way too loud anyway.
    I can't take super accurate distortion measurements anymore because the one WM61A microphone with the lowest distortion died, or at least has higher distortion now and my other mics have much higher distortion. That's probably why you haven't seen measurements with much less than 0.1% distortion - the measurement equipment really starts to matter at a certain point, even more so than the headphones measured.
     
  6. murphythecat

    murphythecat GRU-powered uniformed trumpkin

    Pyrate Banned
    Joined:
    Apr 4, 2016
    Likes Received:
    1,201
    Trophy Points:
    93
    are you sure the 2khz peak is from the voxativ itself?

    can you post measurements of the voxatif in free field not mounted on your baffle?

    at what frequency do you want to cross the voxatif to the woofer?
     
  7. JonnieD

    JonnieD Acquaintance

    Joined:
    May 16, 2017
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    London
    I'm late to this party as I'm new here.

    But I tried series resistance to make higher Q but I don't think it was any better than using lower Q drivers like these: http://aespeakers.com/shop/dipole/dipole15/ as they are good value in my opinion by the time you've padded down lower Q drivers like the PD's (which are good though and cheap here in the UK!).

    Interesting thread though, and good to see others using wideband drivers like mine!

    I mate my AER BD3 / 200hz front horn with a 2 x 15" AE (actually the older Lamba ones) on a straight baffle for bass. And in the end I went (Class D) active on the bass as this gave me the best performance overall. I spent months trying a fully passive system and gave up for a few reasons. 1) I hate passive crossovers, it costs loads and even with a lot of modelling you need to solder and try, 2) I wanted to use low power SE Triode amps, and none are any good in the bass region. 3) I wanted as low a profile system as possible, 6' x 3' baffles are great for OB bass but suck aesthetically!
     
  8. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Yes, the 2kHz peak is there in anechoic conditions. I doubt the baffle changes much. In-room it's very smooth.

    Crossover is somewhere around 300-350Hz depending on the distance to the speakers, but I'm still not 100% done with the crossover. I'll get up measurements of the current version (the one that @Muse Wanderer heard) later.

    That sounds very interesting! Which kind of horn profile do you use? Is it similar to the Oris horns? I kind of like how horns reduce the room reverb and give you much higher sensitivity. Can you still hear the horn coloration?
    What slope is your crossover now? I found that the anechoic measurements for the woofers didn't work for making a crossover that works in-room, so I used smoothed and gated in-room measurement. Even then for tweaking the crossover I measured the FR in-room for the individual drivers and both drivers at once and looked at the CSDs. This works well because I'm essentially using a first order crossover. I looked at the CSDs for the <2kHz region and looked for something that gave me a mostly smooth response <1kHz, with the least amount of cancellation. I get very close to no excess phase in the crossover region and even get a reasonably smooth triangle step response at the listening position. The widebanders with whizzer cones don't quite behave as minimum-phase systems themselves, so a perfect step response isn't possible, but I personally don't really think you can hear it.

    What does your system look like now? Which drivers have you tried?
     
  9. JonnieD

    JonnieD Acquaintance

    Joined:
    May 16, 2017
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    London
    Yes! I'm using Oris 200 front horns with the AER BD3's, and as you'll see below a switchable Beyma tweeter. I don't hear any colouration on this really, that horn was well designed around the AER's and is well integrated and measures surprisingly flat. I'm using first order crossover at 300hz, just a single Teflon cap on the input to my SET amp. So effectively its second order acoustically. And then 4th order (electronically) on the ICEpower bass amps.

    Sounds like you've done a ton of work on this and you've taken it as far as possible. Nice job!

    I used digital crossovers and PC measurement to get my crossover right, and then implemented it as above in the analogue domain (cheapo Behringer DCX and a decent measurement mic and a laptop).

    Like this:

    [​IMG]

    [​IMG]

    In this system the fullrange drivers I tried are most of the Lowther models and the AER drivers (BD1 & BD3), plus a few Fostex, Audax ones too. Of all these the AER's are best by far.

    In the bass I've had Precision Devices, some Eminence (can't remember which but upper end 15"'s with high Q) and two types of Lambda, the Dipoles and the regular TD15's.
     
  10. Poleepkwa

    Poleepkwa Friend

    Pyrate BWC
    Joined:
    Oct 1, 2015
    Likes Received:
    1,557
    Trophy Points:
    93
    Location:
    Finland
  11. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,775
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    ORIS. Yum. This is the next logical step for me. The Altec horns / compression drivers are not resolving enough. Obviously, the next step is to put a good wide bander in a front horn.
     
  12. JonnieD

    JonnieD Acquaintance

    Joined:
    May 16, 2017
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    London
    I'd say go for it!
    There are so many reasons that a "controlled directivity" solution for the mid and high range makes sense, as I'm sure you know. Not "constant directivity" as I think its desirable for that to change as frequency increases. And I agree, though compression drivers do seem like a good solution on the face of it, in practice there's nothing like a decent sized throat being driven by a high quality wide range driver into a circular front horn. A lower throat to mouth ratio is always going to be easier to match.

    I wonder if you can find someone in the States that can make you a horn with a Le Cleac'h flare? I'm sure I've read up on some clever guys that are switched on to this there. But the Le Cleac'h profile that has a wider mouth radius makes for a smoother frequency response and better phase behaviour. I've heard a 204hz Le Cleac'h horn with an AER and its even better than the Oris.
     
  13. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    89,775
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I might just order a pair of ORIS and have them shipped. I used to run a JBL compression driver with a JBL CD horn. They were a bit too much in terms of constant directivity - heh - in the home environment. The Altec multi-cells I'm using now are actually better in the house. My other speaker setup uses the 8" Cicada wide-bander drivers (on occasion, I'll throw in the Lowthers.) I'd really like to get the best of both worlds. The resolution of the wide-banders and the explosiveness / lack-of-compression of the front loaded horns.

    Bruce Edgar (local) made great horns, but I am not sure if he is in good health anymore. He used to have a site with his plans, but he took it down many years ago.
     
  14. spwath

    spwath Hijinks master cum laudle

    Pyrate BWC
    Joined:
    Dec 13, 2015
    Likes Received:
    7,894
    Trophy Points:
    113
    Location:
    Madison, WI
    Ive always admired those, look at this complexity and craftsmanship
    [​IMG]
     
  15. JonnieD

    JonnieD Acquaintance

    Joined:
    May 16, 2017
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    London
    Its an Oris 250 horn now rather than the 200, I don't know how they differ. The 250 may have been designed around the use of compression drivers rather than the 8" wide banders, as that was the direction Bert was headed.

    It might be interesting to see how much it would cost to get a pair of these to the USA http://azurahorn.com/azurahorn_horns.html The AH-204 is better than the Oris 200 due to the Le Cléac’h flare, I've heard my friends ones in London. They're bigger though due to the wide mouth radius.

    Yep, that was my objective too. Here was an earlier experiment, Oris 150 horn on the front of an oversized Cicada like cabinet!

    [​IMG]

    Sounded great, needed EQ as you would imagine, but well over 100db efficiency.

    @Serious - sorry we're hijacking your thread here!
     
  16. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Now I still don't get this. Is the OB null directed at the listening position? Why?
    Are you using copper foil cable for your AERs? I see you're using the phase plug. What did that change?


    Thanks! I still have lots of work to do to get it to work exactly how I want, though.

    First up and most importantly, here is the impedance for the whole speaker. The slight difference in Fs is caused by slightly different U-frame profiles on the left and right side where I just didn't bother to get them perfect. Fs is around 32Hz, without the U frame I measured almost exactly 34.6Hz for all 4 woofers. The difference in the higher frequencies is because the Voxativ driver matching isn't perfect. The matching for the woofers actually seemed better than for the Voxativ drivers, at least the impedance. On-axis the FR is almost exactly the same for the two Voxativ drivers though, but it does divert a bit off-axis. The off-axis response for both drivers is actually slightly different to one side than to the other.
    Impedance.jpg
    The impedance is around 3 Ohm for the whole lower midrange where there is a lot of energy with most music. Probably not very tube amp friendly. But the impedance curve is nicely smooth with no nasty phase angles.
    The crossover I'm using right now is a 1.6mH coil in series and a 136uF cap and 2 Ohm resistor in parallel as a Zobel circuit. The Voxativ driver has no crossover parts.

    Here's an example of one of the CSD measurements I took when I tweaked the crossover:
    XO CSD 1.jpg
    The red graph is just the Voxativ driver, while the blue graph is both drivers. As you can see the woofers actually shaves off about 1.5db off the 2kHz peak. The interesting part here is the range below 1kHz. The CSD isn't really accurate above 1kHz. The CSDs vary a lot with different measuring positions, but overall the range below 1kHz is reasonably smooth.

    The FR at the listening position is a little harder to measure. Lots of floor and wall bounce issues right around the crossover point that I tried to average out a little. I measured from the Gungnir Multibit BAL out with the Ragnarok set to max on low-gain, (which should be unity gain) at -43dbfs and raised it 40db to reference it to 2.83Vrms. I think the level looks a little high to me, but it could just be how the room reflections add to the sensitivity. Sensitivity looks to be around 101db from this measurement. The listening position is about 2m away.
    Avg LR.jpg
    The tuning might still be a little lean, but I already quite like it. I can't get much more bass than this without padding down the widebander, which I do not want to do at all. The only way to get more bass is to either get even bigger baffles (unlikely) or get more membrane area for the lows (I doubt I will get more than two woofers per side, as I feel more would hurt the imaging, but bigger bass drivers are still an option). My baffle is large enough to mount 21" woofers if I wanted.
    I also haven't measured the DC resistance of the coils I'm using. Most likely there's still some room for improvement there, but air core coils with low DCR get expensive. Right now I'm using one air core coil (0.5mH) and an old cored coil that I found (1.1mH).
    The bass actually seems to extend fairly well to the lowest room mode (24Hz), but subjectively I wouldn't say this is the case. At the listening position the graph is almost flat from 35-24Hz, but subjectively I don't get much bass below 30Hz. I guess it's probably because the room mode essentially takes time to fully build up.

    Here's the step response at the listening position:
    Step response R.jpg
    It doesn't look perfect, but that is mainly just how widebanders with whizzer cones measure.
    Here's a gated measurement with the excess phase added. 34ms BH7 window. The excess phase is nicely flat across the crossover range (there is a reflection around 250Hz and obviously more severe reflections higher up), but the excess phase starts to look a little more wild once we approach the upper midrange and also the upper treble. This is what causes the not-quite-perfect step response.
    Phase2.jpg
    I still have to optimize the distance between the drivers so the excess phase response will most likely look even better after that.

    EDIT: Some updates
     
    Last edited: May 25, 2017
  17. JonnieD

    JonnieD Acquaintance

    Joined:
    May 16, 2017
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    London
    No, those pictures are misleading. The bass drivers are around 26 degrees rotated from the front horn, but at the listening when the horns are crossing over in front of the listening position the bass units are virtually facing into the room longitudinally.

    I'm using flat copper foil (similar to Townshend and Goertz) cables for the AER's yes, they sound great.

    The phase plugs in the picture are the BD-Design brass ones, they do seem to smooth out the treble range, it can be measured.
     
  18. JonnieD

    JonnieD Acquaintance

    Joined:
    May 16, 2017
    Likes Received:
    62
    Trophy Points:
    18
    Location:
    London
    It all looks pretty good to me, I'm not sure what else you could change that wouldn't mess up the sound in some way.

    Did you try any different bass drivers, and are those 4 ohm ones you're using? And are the Voxativs still 12 ohms or lower now? The system impedance is lower than I would like to see for myself, but then I do use tube amps. I'm sure SS ones are no problem for yours but given the huge efficiency they'd sound great I'm sure with a nice SET amp or maybe an OTL!

    That's quite a short listening position for that set-up I'd say. So if its as good as that it seems the crossover is working well to get that even a FR.

    The disparity in the Voxativs in the high end is odd, do they measure like that nearfield? I can't see what room effects might affect that, so I wonder if your dirvers are a bit off or whether its typical for the Voxativs. I know that some of Lowther ones are notoriously poorly matched, but the AER's were pretty good. It is hard though for such a high efficiency driver to get good matching, especially with a whizzer cone.

    Did you try different phase plugs too?
     
  19. Serious

    Serious Inquisitive Frequency Response Plot

    Pyrate BWC MZR
    Joined:
    Sep 28, 2015
    Likes Received:
    2,594
    Trophy Points:
    113
    Location:
    near Munich, Germany
    Oh, that makes more sense now.
    Nice, I might want to try copper foil cable myself at one point.

    The crossover still isn't quite where I'd like. I want to get rid of the supports that are in front of the Voxativ baffle - I'm sure they cause some reflection/diffraction issues, although lining them with thin felt didn't seem to make a big difference (could be that the felt was too thin). The overall listening height might still be a bit too high for me. The centre of the Voxativ driver is 92cm off the ground.
    I also want to try what it sounds like with one large baffle instead of 3 smaller ones, but because Voxativ driver baffle essentially is the crossover I doubt it would work. In that case I'd also want to see what it's like with the woofer magnets facing the listening position - that might get reasonably close to time-aligned drivers without a stepped baffle. That might not work since the midrange response of the woofers is important for the right crossover.
    The drivers aren't perfectly time-aligned yet. I want to build something that allows me to move the Voxativ driver baffle back and forth. In terms of measurements a 10cm difference shouldn't make much of a difference for a 1m wavelength (at the crossover region), but it does most likely affect the response above the crossover region. And my gut tells me that precisely time-aligned drivers, even at such low frequencies, make a difference for the imaging.

    Nope, for now only the Eminence Beta 15s. Two 8 Ohm drivers in parallel, otherwise it's impossible to match the sensitivity, especially in an open baffle. The Voxativs are around 10.5 Ohm at 400Hz, which is the lowest point in the audio region, rising to about 25 Ohm at 20kHz.
    Yes, the impedance is quite low. For now I only have the Ragnarok, but I'll have to find a solution for when I get a SET amp at some point in the future. I might want to look into biamping then.

    The nearfield measurements are very close on-axis. Off-axis there's some slight drift. The off-axis response isn't exactly the same to all sides for some reason.
    Here's an older version (with just one woofer) where I simply moved the left speaker to the right and vice versa. Averaged FR of left and right. The individual left and right channels were pretty close, but one side had a peak and the other side didn't. I guess I got lucky that the peak was to the left side of one driver and to the right side of the other driver, so that both don't have the peak at the listening position. LR reversal.png
    As for the LR matching in the "OB in-room" graph, I also didn't make super sure to match the region above 1kHz, as I was mainly interested in the crossover region.

    Here's a picture that should be more indicative of the matching. Averaged both the before and after swapping the speakers for each driver:
    LR matching 1.jpg
    I have no idea where the weird 700Hz spike came from in that measurement, but overall this is more indicative of both the FR above 1kHz and the driver matching than the measurement in the other post. It's still not quite perfect since I only averaged from two measurements and it could be that I moved the microphone between the measurements, but it should give you a good idea. The driver matching is pretty good, but not 100% perfect like you get with a Sennheiser HD800.

    I haven't tried any other phase plugs and I also haven't tried it without the phase plugs.
     
  20. Shinwami

    Shinwami New

    Banned
    Joined:
    Nov 13, 2016
    Likes Received:
    11
    Trophy Points:
    3
    Location:
    Okinawa
    what' the lowest THD 15" baby you could find? finding for years, no luck
     

Share This Page